sharadsinha

Posts Tagged ‘Interdisciplinary research’

Translational Research: What I learned doing (seemingly) mundane task of video annotation

In Design Methodologies, Education, Embedded Systems, Engineering Principles, Research and Development on November 27, 2016 at 3:04 PM

In the recent past I have been doing some work related to automatic video annotation. Videos that you and I take can be annotated with data about the contents of the video. The contents of the video can mean: objects, their types, their shape, background scene (moving or static), number of objects, static and in-motion objects, color of objects etc. One would like to keep a track of objects as the video progresses. Tracking helps in knowing when an object appeared in the scene and when it disappeared. All of the prior work on automatic video annotation is not really completely automatic [1], [2] etc.. They are semi-automatic at best and manual input and control is still required when annotating using these methods.

While doing this work, I developed a better understanding of some of the so called “automatic object tracking for surveillance” solutions out there in the market.  None of these solutions can ensure a complete hands-off scenario for humans. Humans still need to be involved and there are reasons for that.  At the same time, it is also possible to do everything in cloud (including human interaction) and claim it as “hands off for a user”. In this case, it is simply that the client is paying someone else to provide the service. It is not a stand-alone autopilot kind of system installed in a user’s premises. Real automatic video annotation is extremely hard, especially when the scene can change without any guarantees. If we add “video analytics” i.e. ability to analyse the video automatically to detect a certain set of activities, it again becomes very difficult to propose a general solution. So, assumptions are again made and these can be based on user requirements or can be domain specific (say tennis video analytics at Wimbledon). Here is a system which may be of interest to you: IBM’s Digital Video Surveillance Service and a few others described in the paper titled “Automated visual surveillance in realistic scenarios“.

Most of the research work makes certain assumptions either about the scenes or about the methods they use. These assumptions simply fail in real world scenarios. These methods may work under a “restricted real world view” made using a set of assumptions, but when assumptions fail, these methods become limited in applicability.

I believe this is a critical issue that many researchers who want to translate their work into usable products have to understand. This is where both strong theoretical and practical foundations in a discipline are needed: theory gives the methods and the tools, engineering tells you what can/cannot be done and the two can interact back and forth.

Advertisements

The Curious Case of Algorithms

In Education, Interdisciplinary Science, Mathematics on October 31, 2013 at 2:40 PM

I finished reading “The Golden Ticket: P, NP and The Search for The Impossible” some time back. It is a very nice book that introduces one to complexity theory. Essentially, it describes, without too much of Mathematics, what kinds of problems can be solved and what other kinds will take forever to solve. However, if these – the forever to solve ones– were to get solved one day, what would be the impact. P  refers to the problems that can be solved quickly using computers to get the best solution. On the other hand NP refers to problems whose best solution cannot be found quickly using computers. I have deliberately simplified things for your understanding. This field is vastly complex!

The word “quickly” is used here with reference to a time span which is acceptable to the seeker of the solution. It could be a few seconds, or a few weeks.  Going by the nature of humans, any solution (best or otherwise) that might be delivered in months or years will probably be unacceptable. The search of any solution is accomplished using algorithms. It is these algorithms that can either give us a solution “quickly” or might take ages to finish their task. It is believed that if we could find algorithms that could solve any problem in the class of NP problemswe could solve many challenges facing us. These problems can be found in varied fields like biology, cancer research, mathematics, computer science, economics etc. However, some of the modern day systems which we feel very secure and safe about will lose these strengths if an NP problem is solved. This is because they rely on the fact that NP problems are extremely hard to solve quickly. For instance, your secure online bank transaction won’t be secure anymore. The public-key cryptography, on which it relies, would be broken by then.

Another technologically interesting aspect of algorithms is their ability to provide information based on someone’s taste in color, clothes, books, music etc. In fact, it is this type of algorithms which is used by eBay, Amazon etc. to recommend to users items for purchase. They track their actions: which items they click on, which items they buy etc. to create an “algorithmic profile” of users. While all this sounds interesting and potentially time saving for someone who knows what to buy, this also has a negative side effect. As a regular user of such platforms, you end up getting information that is tailored to your existing taste. Therefore, you cannot easily get information that is not relevant to your taste. Effectively, your ability to explore ( if you are also someone who likes to explore) becomes limited. Of course there are ways to overcome this, simplest of them being not to sign in when performing a search!! You can argue that many prefer automatic sign-ins to save time and the need to remember passwords. True, but then you have to decide whether you want to work/live like a frog in a well or like a whale exploring an ocean! 🙂

Technology Innovation and Unemployment

In Education, Interdisciplinary Science, Research and Development on September 30, 2013 at 12:48 AM

Automation has increased productivity in many areas. If you look at the assembly line or shop floor of a car manufacturer, you will see automation in its full might. Though you will still find a certain number of workers, their number is far less compared to pre-automation days. You may have also come across call center staff who deal with queries related to insurance, bank related tasks etc. Most of these queries are routine in nature and it is the same kind of information that the staff has to provide to the callers. There is recent news that companies like IPsoft are providing artificial intelligence based virtual call center staff to handle such queries. This is expected to reduce the number of people required in BPOs and call centers. 

An aprocryphal tale is about a conversation between Henry Ford II and Walter Reuther. The former was the head of Ford Motor Company while the latter controlled its union. When Ford asked Reuther how he would make robots pay union dues, Reuther asked in return if Ford could make his robots buy cars. Ford got the point that any increase in productivity has to be met with an increase in the number of consumers. Ford raised the salary of his staff so that they could afford to buy cars. 

Do you think that an increasing rate of technological innovation can lead to rise in unemployment? If you believe in this, you probably believe in Luddite Fallacy. I would rather suggest to be open to debates on this issue. This issue is far from resolved and new insights keep coming now and then. Two opposing views on this issue can be found here and here published in The Economist and Forbes respectively.

Given the fact that many engineers work on systems which are meant to increase productivity, provide better services, it is only relevant to have a look at an aspect of economics and social change that they are seldom concerned with. It is not so much about questioning what they do rather it is more about understanding the mysterious ways in which the world moves!

Doers vs.Thinkers or Doer-Thinker or Thinker-Doer

In Education, Research and Development, Startup on July 14, 2013 at 7:09 PM

Have you ever found yourself thinking about the topic of this post? Quite likely. And it is all the more likely that you tried to classify yourself as a doer or a thinker. Being human beings, we love to take sides most of the time. We (or others) are either this or that; we work this way or that way etc. Doers take pride in doing things while thinkers take pride in their ability to think deeply and profoundly. Many doers challenge thinkers by saying that they “only think” and do very little on the ground while many thinkers hold the ability to think and come up with ideas in the highest regard. A classic case is the question about work experience versus research experience which I discussed here.

If we stop for a moment and decide not to analyze these two traits from a usability perspective and detach them from their economic outcomes, what do you think will win the doer vs. thinker debate? I guess it will still remain open because in their absolute existence, these two traits offer two different kinds of results. While the former generally gives birth to something tangible, something that our sensory perceptions can respond to, the latter gives birth to something that our minds can (or cannot) comprehend. For instance, a carpenter can produce chairs, tables etc. while someone who studies trees and plants can propose a theory on growth of trees. These two guys can exist in isolation without any problems. Problems arise when we try to assign a monetary value to their efforts and it is then that the debate starts. Since the aim of practically all economic exercises is to maximize the return on money and time invested, whether a doer is more important or a thinker, depends on who brings more value in a given context.

However, instead of identifying yourself with one of these, you can as well be both of these: doer-thinker or thinker-doer. You are one of these two depending on the ratio of these two traits in your character. The good thing about being both is that you appreciate both. You do not become dogmatic and you understand the effort and the skills required for each of them. You can appreciate both kinds of people (who exist in the either or world). Your attitude, character and style of functioning becomes more fluid and you probably gain the knowledge to get  the best out of not only yourself but also out of those who exist in silos. All this becomes really helpful when you are in an organization or you are leading a team. Thinkers can inject new and fresh ideas while doers can execute them. But as you are both, you know very well that an interaction between these two may lead to even better results than the sum of their individual results.

Soilless Farming and “Re”search

In Education, Engineering Principles, Research and Development on June 25, 2013 at 12:12 AM

When I started my PhD, my supervisor, among other things, told me that research is also revisiting the existing concepts and examining them. It is not always about plucking a blue-sky idea from nowhere or dreaming up something like that out of nowhere. That is why it is called “re”search. Based on my experience over the last few years, I now firmly believe in what he said. Very often we try to come up with an idea that we want to sound extraordinary. We want to come up with something that inspires awe and gaze. Nothing wrong in that, except that looking at the history of technological evolution, it can be seen that ideas and technologies that have been considered ground breaking and have held us in thrall, have often come up revisiting the existing concepts. Of course there are those which were the results of serendipity, for instance the discovery of penicillin. But that is not the topic of this post.

By examining closely what is considered common knowledge or given fact, people have made breakthroughs. Agriculture has long been associated with soil based farming. In fact, we seldom talk about agriculture without associating quality of soil with it. Agriculture, as we have known over thousands of years, cannot be practised without soil. However, Dr. Yuichi Mori, a professor in Japan, has re-examined the role of soil and realized that soil can be replaced by a suitable membrane that can provide nutrients to plants and physical support for roots to grow. This is “soil-less agriculture“. His company Mebiol markets the technology called Imec. Not only the technology does not need soil, the hydroponic membrane stores water and nutrients leading to need for less water for plant growth. The membrane may also block some pathogens that cause plant diseases. Field trials have shown that tomatoes, cucumber etc. can be easily grown and grown this way they in fact taste better and richer in nutrients. You can watch his TEDxTokyo talk here.

Amazing, isn’t it? Now I can safely try to grow some of these if I were to live in a land scarce country or in a high rise apartment! Interestingly, the earliest documentary proof of the idea of soil-less agriculture can be found in 1627 book Sylva Sylvarum by Francis Bacon with follow up research by some people over the next few centuries. However, Mebiol is the first company to come up with a technology that can be commercialized.

On Diffusion of Innovations

In Education, Interdisciplinary Science, Research and Development, Startup on May 10, 2013 at 1:57 AM

Diffusion of Innovations is a remarkable book by Everett M. Rogers. It is also a field of study and research where questions related to the diffusion of innovations through different groups of people and cultures are studied. This theory seeks to explain how innovations spread, how they are adopted or rejected, their social impact and the rate at which these processes occur over a period of time. This book has plenty of examples of innovations that diffused and those that did not. Notable examples include the idea of water boiling that the public health service in Peru wanted to promote in a Peruvian village and failed in doing so; non-diffusion of the Dvorak keyboard; the relatively successful STOP AIDS campaign in San Francisco in the mid-1980s etc. Note that the use of the term innovation  is not restricted to technological innovations only. According to Rogers, “An  innovation is an idea, practice, or object that is perceived as new by an individual or other unit of adoption“.

Technologists and engineers generally think that a new idea will sell itself, that advantageous innovations will be quickly adopted. However, this is seldom the case and the adoption, in general, is slow. This is a fact that is of relevance to many start ups. There are social and cultural aspects of innovation that have a big influence on its adoption. Influencing the adopters involves not only relevant marketing but also addressing social, cultural and economic issues. Of course the range of issues to be addressed depends on the innovation that we are trying to sell or promote.

It would come as a surprise to many that Everett M. Rogers was not from business or engineering background. He was a scholar in  communications and sociology!

Complementary to Google Search

In Education, Interdisciplinary Science, Research and Development on March 26, 2013 at 5:31 PM

When we look for information, almost all  of us invariably turn to Google. There is no doubt that Google and its services, especially its search engine, have helped many of us who look for both new and old pieces of information often. However, is it always the best in terms of returning results that are relevant to the search query? Not necessarily. By relevant, I mean the intention of the user who typed in the search query.  Given the fact that the Google page rank algorithm, among other things, assesses the importance of a page/resource based on the number of pages/resources linking to it and their importance as well, the search result tends to tilt in favor of  those resource which most people are talking about. Therefore the search result can include Wikipedia references, journal and magazine articles, Youtube videos etc. This means that there may not be a uniformly decreasing order of depth of information available in the search results. It may also mean that the quality and expanse of information available in the search results could vary widely. For instance, a news report that shows up higher on the list of search results could be discussing the outbreak of a particular disease and its economic and social impacts. It may not be discussing the medical science behind the disease itself. Of course one can try combinations of phrases as well as Google custom search to narrow down the results. This also means that if you are searching for something that is discussed rarely, you will have to sift through a lot of results. However, there is another way of looking for targeted information: use of field specific search engines. For example,  FindZebra  is a search engine for rare/orphan diseases. This is currently a research project at Technical University of Denmark  and it seeks to help doctors looking for information on such diseases for the purpose of medical diagnosis (an example of very targeted information). It indexes only the most relevant databases for this purpose.  Its comparison with Google search and Google custom search can be found here. Archives, like this one,  in different fields also serve the same purpose. It  will be better to see the two search approaches as complimentary to each other. While field specific search engines/archives can be very precise, Google search can provide a wider set of results where different perspectives and analyses may emanate on the same subject but from  people with different backgrounds.

How much and what do you read as a researcher?

In Education, Engineering Principles, Interdisciplinary Science, Research and Development on March 3, 2013 at 5:07 PM

What do you read as a researcher? Most of us read only that which is relevant (or we think is relevant) to our research. But is that all that is should be read? I know that many of us do read novels of different kinds of which fiction is more common.

However, as far as reading for research is concerned, most of us read within our specific domain and especially focusing on those works that are closely related to our own. We browse through conference proceedings and journals a lot. Some of us venture into reading patents and online newsletters published like EE Times  etc. Nevertheless we tend to stick to a rather narrow range of topics. We measure the utility of reading something for research against the value that it might bring, in our opinion, to our research. While this is not at all a bad way of doing research, we run the risk of training ourselves to read, think and argue about only a very narrow set of topics even within our own broader research discipline. It is a byproduct that has its negative consequences. It becomes difficult to think beyond what we are most comfortable with and it makes an expert in a very narrow field. We run the risk of not being able to relate our work with the bigger picture and processes. We run the risk of not being able to think at the system level or looking at the same thing from a different perspective. For instance, a mobile phone is a device that has both software and hardware. A software guy will describe it from software perspective while the hardware guy from hardware. Someone who can understand both, even if not every detail, can help merge the two perspectives which is very important for product design!

Oscar Wilde has said, “It is what you read when you don’t have to that determines what you will be when you can’t help it“. It applies not only to life but also to research. Reading about human factors, user interfaces, intellectual property, regulatory practices etc. helps us in seeing same things from different perspectives. It is a great way to exercise our brains.

At the same time,if you are more adventurous , reading about topics in sociology, psychology, economics, politics etc. helps you develop critical thinking abilities borrowed from different domains. An example is here. And if you can see through all of this, you might even be able to solve a problem in your domain by reading about something exciting in another domain.

Value of Scientific Work

In Education, Interdisciplinary Science, Research and Development on February 21, 2013 at 11:03 PM

We, researchers and non-researchers alike, often come across this question: what is the value of scientific work? Is it about publishing papers that expound upon novel ideas? Is it about work that can be commercialized? Is theoretical scientific work more valuable than practical scientific work? Are hands on abilities more valuable than critical thinking? These are perplexing questions with no straight forward answer. Probably, the best way to think of these is to understand that there is space in this world for all. Different approaches have their pros and cons. I think that it is even better to answer them in some context rather than after untying them from any context.

I have seen graduate students, especially PhD students, grappling with the value of their work in the beginning and then towards the end of their PhD. Many find their work of not much value compared to work that translates into something tangible, something that you can touch or see or hear; something that people can use; something that makes them feel that they have created something that did not exist.

Nevertheless, the use of an idea and the idea itself are two different things. They each deserve their own attention. It is also possible that a use case for an idea may not be realized in the immediate future once the idea was formulated. It may take time for the use case to appear and it may not be the same person who developed the idea. For instance, the inventor of laser would have probably never imagined that one day it would be used in laser pointers which we often use during presentations.

I think that when we question ourselves in this way, we need to go back in time and see what the savants of the field have said about these. One great piece is titled “The Value of Science” by Richard P.  Feynman, Physics Nobel Laureate (1965) and both a researcher and a teacher par excellence.  It is so rare to find an excellent researcher who is also an excellent teacher these days. The following paragraph from his public speech will probably be of immense interest to young researchers:

“I would now like to turn to a third value that science has.  It is a little less direct, but not much.  The scientist has a lot of experience with ignorance and doubt and uncertainty, and this experience is of very great importance, I think. …………….  We have found it of paramount importance that in order to progress we must recognize our ignorance and leave room for doubt.  Scientific knowledge is a body of statements of varying degrees of certainty – some most unsure, some nearly sure, but none absolutely certain.”

For a funny and amusing reflection on the different kinds of researchers today ;), see this by Daniel Lemire, a  computer scientist with a big following.

Role of Industrial Consortia in Education and Research

In Education, Embedded Systems, Industrial Consortia, Research and Development on February 8, 2013 at 6:58 PM

A Google search will reveal the existence of quite a few influential industrial consortia further the cause of research and education in fields identified by them. Almost all of them are run jointly by people from industry and prominent educational and research institutions. You can find a list of them compiled here. I have listed only the ones relevant to electronics and computer industries. I have found that not many students are aware of these consortia and that should not be the case. Some of these are highly active and they contribute a lot to research, development of technology and education. Consortia like Accellera Systems Intiative have contributed to a number of IEEE standards. Some of these can be downloaded for free from its website. The Semiconductor Research Association plays an important role in promoting research and education in the field of semiconductors. The International Technology Roadmap for Semiconductors has played an immense role in identifying challenges before the semiconductor industry- from design to manufacturing, to testing and validation. Many of these associations also offer scholarships and fellowships for students and research grants for faculty members. Their publications provide a lot of insight regarding the challenges at present and of the future. These publications may not always have a lot of in depth research material, the sort of which most graduate students are accustomed to, but they successfully paint the bigger picture. Paying attention to such facts can help in keeping research relevant to industry where necessary. Besides, it also helps in learning about the actual real world problems and the challenges involved in translating research into technology that can be scaled up and widely used. Sometimes, problems are considered solved in academic research but such solutions never make it to the market, even if of relevance, because their translation to scalable technology still remains an open problem.